Table 5 Execution time(s) of GPU kernels with different mesh loop modes on V100
From: Effects of mesh loop modes on performance of unstructured finite volume GPU simulations
Kernels | Mesh 1 | Mesh 2 | Mesh 3 | Mesh 4 | Mesh 5 |
|---|---|---|---|---|---|
NV-F | 0.460 | 1.307 | 3.048 | 5.758 | 16.940 |
NV-C | 0.122 | 0.480 | 1.243 | 2.462 | 6.613 |
NV-N | 0.393 | 0.976 | 1.867 | 3.005 | 5.047 |
GQ-F | 0.061 | 0.154 | 0.308 | 0.537 | 1.209 |
GQ-C | 0.134 | 0.326 | 0.639 | 1.102 | 2.198 |
GQ-N | 1.361 | 3.354 | 6.812 | 10.579 | 19.151 |
SF-F | 0.099 | 0.252 | 0.513 | 0.898 | 1.983 |
SF-C | 0.149 | 0.343 | 0.638 | 1.057 | 1.856 |
LM-F | 0.050 | 0.116 | 0.229 | 0.389 | 0.834 |
LM-C | 0.020 | 0.048 | 0.094 | 0.170 | 0.362 |